A multilevel statistical technique to identify the dominant landscape metrics of greenspace for determining land surface temperature

Although considerable effort has been made to investigate the complex deriving forces of land surface temperature (LST), relatively little attention has been paid to the issue of spatial hierarchy, which is an intrinsic property of the urban system. To solve this problem, we propose a multilevel statistical technique, incorporating a regression tree and an improved hierarchical partitioning model, to investigate the hierarchical effects of greenspace spatial patterns on LST. We tested the technique in Guangzhou with two Landsat 5 images acquired in summer. The results show that the proposed technique explicitly identifies the nested hierarchical structure of LST. Greenspace spatial composition is the dominant metric at the highest level, whereas some spatial configuration metrics affect LST variations more in the lower levels. One obvious advantage of the technique is its ability to identify the dominant landscape metrics of greenspace and to determine the hierarchical effects of LST formation. The technique serves as an important approach for environmental studies of urban heat, and it furthers our understanding of the complex and hierarchical mechanism of LST patterns and processes in urban areas. The idea of this multilevel statistical technique is also applicable to the study of other mechanisms in urban systems.